The detail in one row that rewrote the others

We found this on our own site this morning, on the least glamorous page we own. Our privacy policy lists the companies that process things on our behalf, one bullet each. One of those bullets had just been extended, carefully, to say that photos go to that provider directly and are never made publicly reachable, and that one particular file travels instead as a private link that expires within the hour. Every word of it was true.

The bullet immediately above it names a different provider and says, in effect, that it reads your images. It does not say how. It never had to, because until that morning no bullet on the page had raised the question. Now one had, and the answer for the quiet bullet turned out to be the interesting one: that provider is handed a link too. The page had not become false. It had become a page where a reader could work out the wrong answer from what was in front of them, which is worse than it sounds, because they would have been reading carefully to do it.

Why the detailed row changes the quiet ones

A reader does not take each row of a table on its own terms. They infer a standard from the rows that are filled in, and apply it to the rows that are not. This is not careless reading, it is the only economical way to read a table, and it is why a table is useful at all.

So when one row answers a question none of the others answer, exactly two readings are available, and both of them are yours to live with:

• The detail is the house standard, and the quiet rows meet it too. The reader assumes every provider gets photos directly, every variant is nickel-free, every size is cut from the same cloth. If that is not so, you have made a claim you never wrote down.
• The detail is remarkable, and the quiet rows are different. The reader assumes you spelled it out here precisely because here it is unusual - and reads the silence everywhere else as a quiet no. If the quiet rows are in fact the same, you have just talked a careful reader out of something true.

Notice that you do not get to pick which reading happens. It depends on the reader, and on which way round is worse for you it will often be the wrong one.

Where this sits in a shop

It is not a privacy-policy problem. It is a problem of any set of rows a reader scans side by side, and shops are built out of those.

• Spec tables. One ring says “925 sterling silver, nickel-free” because a customer once asked. The other eleven say “silver”. Nothing about the eleven has changed and the page now implies something about all of them.
• Care instructions. One garment has a full washing paragraph; the rest have a symbol. A reader concludes the rest are less delicate, or that you have not checked.
• Delivery and returns on individual products. A single line on one listing saying returns are accepted unworn within thirty days makes every listing without that line look like a listing with a different policy.
• Materials and origin. “Made in Portugal” on four of nine products is read as “and the other five are not” even when the other five simply have not been filled in yet.
• Sizing notes. One product carries “runs small, size up”. On a page of twenty, that is now the only product a reader believes they have sizing information about.

The common thread is that each of these was added by someone doing the right thing. Nobody writes a careful sentence in order to mislead; they write it because that row genuinely needed it that day. The cost lands somewhere they were not looking.

The check takes ten minutes

Open any page that puts several of your things side by side and read it column by column rather than row by row. For each column, ask which cells are filled and which are empty. You are looking for a column where one or two cells carry a careful, specific sentence and the rest carry nothing. That column is the one doing the damage.

Then ask the question that actually matters, which is not “is each row true”. It is: “if a reader assumed every other row matched this one, would they be right?” If yes, say so in the other rows. If no, you have found something you were accidentally implying.

The fix is not more detail, and it is not less

The obvious response is to delete the careful sentence, because the page was consistent before it arrived. That is the wrong move and it is worth being plain about why: the sentence was added because it was the most useful thing on the page. Removing it restores the consistency by removing the information, which is how pages end up uniformly vague. Shops do this constantly and it reads as having nothing to say.

The other obvious response is to fill in every cell of every column to the same depth. That is better, and it is usually too much work to actually happen, which means in practice it is the same as doing nothing.

What works is narrower. Pick the columns where a reader would make a decision differently depending on the answer - material, care, origin, returns, what happens to something they give you - and bring those columns up to one standard across every row. Leave the columns nobody decides on alone, at whatever depth they are. One column finished beats five columns half-done, because a half-done column is exactly the shape that invites the wrong inference.

We took the first route on our own page: the quiet bullet now says how that provider gets your photos, in the same terms the detailed one uses. It is one sentence. It took longer to notice than to write, which is the usual ratio for this kind of thing.

This is a cousin of something we have written about before. A count in your copy goes stale because nothing is holding it to the set it counts; this goes wrong for the opposite reason - every sentence stays true, and the meaning of the page changes anyway. Both are cases where a promise in your copy is harder to check than a number in it, because what went wrong is not in any one sentence.